Phi-Bench - Inference and Serving: leaderboard
Metric: Infrastructure Score (%). Source: arxiv.org. Saturation forecast: Around March 2028. 8 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Kimi K3 | 30.5 |
| 2 | Qwen 3.8 Max | 29.9 |
| 3 | Claude Opus 5 | 26.2 |
| 4 | GPT-5.6 Sol | 21.8 |
| 5 | GLM-5.2 | 19.2 |
| 6 | Qwen 3.7 Max | 12.7 |
| 7 | Claude Sonnet 5 | 7.2 |
| 8 | DeepSeek V4 Pro | 3.9 |
Interactive version: theaggregate.ai/benchmark?slug=phi-bench-inference-and-serving · How It Works · Data refreshed daily, snapshot 2026-09-24.